# Canvas on AWS **URL:** \ **Category:** Elastic Cloud Enterprise (ECE)\ **Tags:** canvas\ **Created:** [March 7, 2019, 11:30pm UTC](https://discuss.elastic.co/t/canvas-on-aws/171399 "2019-03-07T23:30:37Z")\ **Posts on this page:** 10\ **Page:** 1
**Author:** ![Randy-312](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/randy-312/32/19451_2.png) [@Randy-312](https://discuss.elastic.co/u/Randy-312)\ **Post date:** [March 7, 2019, 11:30pm UTC](https://discuss.elastic.co/t/canvas-on-aws/171399/1 "2019-03-07T23:30:37Z")
Has anyone implemented canvas on AWS? with ECE? We are trying to get it to work and running into some interesting web socket issues, which may be related to our use of ALBs. ---
**Author:** ![zanbel](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/zanbel/32/20678_2.png) [@zanbel](https://discuss.elastic.co/u/zanbel)\ **Post date:** [March 8, 2019, 11:21am UTC](https://discuss.elastic.co/t/canvas-on-aws/171399/2 "2019-03-08T11:21:08Z")
Hey @Randy-312, While ECE proxy layer doesn't support websockets at the moment, Canvas should be able to use polling. Internally we test ECE on AWS among other environments so I think you are correct in the assumption it's related to the configuration on the ALB. Can you share more about the error you are seeing and you ECE version? Would also suggest to post the question in Kibana discuss forum as I suspect this issue might be more generic and not related to ECE, and would happen in self-managed Elasticsearch and Kibana. ---
**Author:** ![Randy-312](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/randy-312/32/19451_2.png) [@Randy-312](https://discuss.elastic.co/u/Randy-312)\ **Post date:** [March 8, 2019, 3:43pm UTC](https://discuss.elastic.co/t/canvas-on-aws/171399/3 "2019-03-08T15:43:23Z")
ece 201 and elastic stack 654. Also tested this on a 3 master 3 data node cluster and canvas loads in 1.1 minutes. At least when there is no insert traffic. It seems to be a problem with our larger clusters though. We have a 50 node cluster and I just gave up after an hour. ---
**Author:** ![zanbel](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/zanbel/32/20678_2.png) [@zanbel](https://discuss.elastic.co/u/zanbel)\ **Post date:** [March 10, 2019, 3:14pm UTC](https://discuss.elastic.co/t/canvas-on-aws/171399/4 "2019-03-10T15:14:33Z")
So the issue is not failing to load canvas, but the time it take to load? ---
**Author:** ![Randy-312](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/randy-312/32/19451_2.png) [@Randy-312](https://discuss.elastic.co/u/Randy-312)\ **Post date:** [March 10, 2019, 3:45pm UTC](https://discuss.elastic.co/t/canvas-on-aws/171399/5 "2019-03-10T15:45:46Z")
It seems to be caught in a nearly endless retry loop, related to cluster size. a 3 node cluster only requiring 1 retry. To me it looks like there is one initial call and then three to four subsequent calls that assume it's connecting to the same original server. This would work well in a websocket protocol however as I pointed out that's turned off for us. aws and ece with ssl. The how many tries before success appears to be (cluster size) ^ 4. I believe there may be a timeout factor here too. Clusters receiving traffic just never come back above the nodes. those without traffic can.. after 1000+ retries. ---
**Author:** ![zanbel](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/zanbel/32/20678_2.png) [@zanbel](https://discuss.elastic.co/u/zanbel)\ **Post date:** [March 11, 2019, 3:40pm UTC](https://discuss.elastic.co/t/canvas-on-aws/171399/6 "2019-03-11T15:40:03Z")
Checked with the team and it will be great if you can open a GH issue in the relevant Kibana repo [https://github.com/elastic/kibana/issues](https://github.com/elastic/kibana/issues). This doesn't seem like an ECE issue so better to move the discussion to GH. ---
**Author:** ![Randy-312](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/randy-312/32/19451_2.png) [@Randy-312](https://discuss.elastic.co/u/Randy-312)\ **Post date:** [March 11, 2019, 7:48pm UTC](https://discuss.elastic.co/t/canvas-on-aws/171399/7 "2019-03-11T19:48:54Z")
> ---
**Author:** ![Randy-312](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/randy-312/32/19451_2.png) [@Randy-312](https://discuss.elastic.co/u/Randy-312)\ **Post date:** [March 13, 2019, 3:30am UTC](https://discuss.elastic.co/t/canvas-on-aws/171399/8 "2019-03-13T03:30:09Z")
So.. may be another issue involved, but we appear to have a workaround, which is showing SOME improvements, although NOT consistently fixing the isssue. I Renamed the Canvas space back to 'Default', and I can now load Canvas. Why did change it from Default in the first place? Since we have Multiple clusters in ECE, and cannot modify their skin or colors, it is difficult to remember what Cluster we are logged into. So, renaming the Kibana space address that. We weren't aware it was going to cause a Canvas impact though. ---
**Author:** ![zanbel](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/zanbel/32/20678_2.png) [@zanbel](https://discuss.elastic.co/u/zanbel)\ **Post date:** [March 13, 2019, 5:05am UTC](https://discuss.elastic.co/t/canvas-on-aws/171399/9 "2019-03-13T05:05:44Z")
Thanks for the update on this thread as well. ---
**Author:** ![system](https://us1.discourse-cdn.com/elastic/original/3X/1/a/1ac57faf039f6b580b3f104ef42a2a89e41014de.png) [@system](https://discuss.elastic.co/u/system)\ **Post date:** [March 27, 2019, 5:05am UTC](https://discuss.elastic.co/t/canvas-on-aws/171399/10 "2019-03-27T05:05:48Z")
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.